Interactive Map

2018/19 Individual Schools Budget Analysis. Data Source = Stats Wales

Please note that a small amount of random noise has been added to school location coordinates in order to avoid overplotting schools that share a campus

Primaries

Primary Phase: Individual Schools Budget vs Pupil Numbers

Linear Regression Analysis

# A tibble: 2 x 7
  Phase   term          estimate std.error statistic  p.value p.adjusted
  <fct>   <chr>            <dbl>     <dbl>     <dbl>    <dbl>      <dbl>
1 Primary (Intercept)      82.9     6.04        13.7 4.34e-40   3.47e-39
2 Primary Pupil_Numbers     3.20    0.0247     130.  0.         0.      

The regression shows that for the primary phase in Welsh schools, each individual student increases the Individual Schools Budget by £3,200. The p value returns a strong statistical significance (p adjusted = 0). Essentially, in 99.9% of cases, an increase in ISB is explained by an increase in pupil numbers. The standard error is extremely low, at a value of 0.0247. This means that on average, the model will be incorrect by £25 plus or minus the median value.

Note that the nursery phase returned a non-significant p value and therefore was not analysed. This is likely due to the small number of nursery schools within Wales.

Interactive Data Table

Secondaries

Secondary Phase: Individual Schools Budget vs Pupil Numbers

Linear Regression Analysis

# A tibble: 2 x 7
  Phase     term          estimate std.error statistic   p.value p.adjusted
  <fct>     <chr>            <dbl>     <dbl>     <dbl>     <dbl>      <dbl>
1 Secondary (Intercept)     434.     66.2         6.57 4.51e- 10  2.25e-  9
2 Secondary Pupil_Numbers     4.15    0.0701     59.3  1.53e-127  1.38e-126

The regression shows that for the secondary phase in Welsh schools, each individual student increases the Individual Schools Budget by £4,150. The p value returns a strong statistical significance (p adjusted = 1.38e-126). In 99.9% of cases, an increase in ISB is explained by an increase in pupil numbers. The standard error is extremely low for this phase, at a value of 0.0701. This means that on average, the model will be incorrect by £70 plus or minus the median value.

Interactive Data Table

Middles

Middle Phase: Individual Schools Budget vs Pupil Numbers

Linear Regression Analysis

# A tibble: 2 x 7
  Phase  term          estimate std.error statistic  p.value p.adjusted
  <fct>  <chr>            <dbl>     <dbl>     <dbl>    <dbl>      <dbl>
1 Middle (Intercept)     168.     196.        0.855 4.04e- 1   6.01e- 1
2 Middle Pupil_Numbers     4.39     0.224    19.6   4.22e-13   2.53e-12

The regression shows that for the middle phase in Welsh schools, each individual student increases the Individual Schools Budget by £4,390. The p value returns a strong statistical significance (p adjusted = 2.53e-12). In 99.9% of cases, an increase in ISB is explained by an increase in pupil numbers. The standard error is very low, at a value of 0.24. This means that on average, the model will be incorrect by £224 plus or minus the median value.

Interactive Data Table

Specials

Special Phase: Individual Schools Budget vs Pupil Numbers

Linear Regression Analysis

# A tibble: 2 x 7
  Phase   term          estimate std.error statistic  p.value p.adjusted
  <fct>   <chr>            <dbl>     <dbl>     <dbl>    <dbl>      <dbl>
1 Special (Intercept)      228.     218.        1.05 3.01e- 1   6.01e- 1
2 Special Pupil_Numbers     18.3      1.63     11.2  9.48e-14   6.63e-13

The regression shows that for the special phase in Welsh schools, each individual student increases the Individual Schools Budget by £18,300. The p value returns a strong statistical significance (p adjusted = 6.63e-13). In 99.9% of cases, an increase in ISB is explained by an increase in pupil numbers.

However, the standard error is relatively larger than the other phases in this report, at a value of 1.63. This means that on average, the model will be incorrect by £1,630 plus or minus the median value. This increase in standard error is symptomatic of having a relatively small sample size in relation to the correlation coefficient and therefore would advise caution in drawing conclusions from this dataset.

Interactive Data Table